Jump to content


0
[Req]

Anyone has the vMX 16.2R2 just released in July?



38 replies to this topic

#1 qianxingnan

qianxingnan

    Junior Member

  • Members
  • PipPip
  • 5 posts
  • 104 thanks

Posted 23 July 2017 - 07:51 PM

I can provide a shared FTP site for upload and download.

#2 HermanToothrot

HermanToothrot

    Member

  • Members
  • PipPip
  • 32 posts
  • 17378 thanks
  • LocationMonkey Island

Posted 23 July 2017 - 10:36 PM

View Postqianxingnan, on 23 July 2017 - 07:51 PM, said:

I can provide a shared FTP site for upload and download.

I ditched it a while back for 17.2R1, but I can download it again, if you provide an FTP server that can handle SCP/SFTP. You want the ESXi or KVM version?


EDIT:  Lemme know all the details before I start downloading whichever version you need. They're both +3GB each.  The ESXi version can actually be shrunk a little, since a good chunk of the size is an upgrade archive.  The 4 required KVM files in the images folder (assuming you'll be using this in EVE or GNS3) will still add up to 3GB.

Edited by HermanToothrot, 23 July 2017 - 10:53 PM.


#3 qianxingnan

qianxingnan

    Junior Member

  • Members
  • PipPip
  • 5 posts
  • 104 thanks

Posted 24 July 2017 - 01:36 AM

Thanks Herman. I already have 17.2R1 but that release seems to have some issues. I keep getting these messages on console:

PE1 fpc0 SCHED: Thread 68 (PFE Statistics) ran for 1030 ms without yielding

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Scheduler Oinker

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 0: sp = 0xda331928, pc = 0x8077095

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 1: sp = 0xda331968, pc = 0x810778b

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 2: sp = 0xda3319a8, pc = 0x811e0d6

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 3: sp = 0xda331e88, pc = 0x8160516

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 4: sp = 0xda331ff8, pc = 0x807cc56

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 5: sp = 0xd7621fd8, pc = 0x8073d35

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 6: sp = 0xd7621ff8, pc = 0x807cc56

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 7: sp = 0xff8254a8, pc = 0x807596a

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 8: sp = 0xff8254c8, pc = 0x807c415

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 9: sp = 0xff8254e8, pc = 0x8064143

Message from [email protected] at Jul 24 01:35:22  ...
PE1 fpc0 Frame 10: sp = 0xff825508, pc = 0x4bf9c9d3


Daniel

Thanked by 1 Member:
kumaragurum

#4 HermanToothrot

HermanToothrot

    Member

  • Members
  • PipPip
  • 32 posts
  • 17378 thanks
  • LocationMonkey Island

Posted 24 July 2017 - 02:45 AM

I'm used to seeing that with the old Firefly Perimeter image (I can't get any of the 15.x vSRX images to work).

Have you tried setting the vFPC to lite-mode?  You'll need to boot all the way into it, and then I think the command is "set chassis fpc 0 pic 0 lite-mode".



But I've never seen those messages with any of the vMX or vQFX images I've tried. Just the vSRX 12.1X47-D20 image.

Edited by HermanToothrot, 24 July 2017 - 02:47 AM.


#5 qianxingnan

qianxingnan

    Junior Member

  • Members
  • PipPip
  • 5 posts
  • 104 thanks

Posted 24 July 2017 - 03:40 AM

Setting fpc to lite-mode didn't fix it. I only see these messages on vMX 17 but not on those old engineering releases.

I am actually trying to lab out multicast MDT but it seems the feature is not available on the engineering releases. That is why I want a vMX 16 to try it out.

Thanks,
Daniel

#6 HermanToothrot

HermanToothrot

    Member

  • Members
  • PipPip
  • 32 posts
  • 17378 thanks
  • LocationMonkey Island

Posted 24 July 2017 - 08:12 PM

View Postqianxingnan, on 24 July 2017 - 03:40 AM, said:

Setting fpc to lite-mode didn't fix it. I only see these messages on vMX 17 but not on those old engineering releases.

I am actually trying to lab out multicast MDT but it seems the feature is not available on the engineering releases. That is why I want a vMX 16 to try it out.

Thanks,
Daniel

Let me know if you want the KVM or ESXi version, and as long as your FTP server will let me use SCP/SFTP (I like using SecureFX), I can download the appropriate file from Juniper, and then upload it to your site.

Worst case, I could upload it to a filehost I know, that doesn't care what size the files are, but you'll want to create a burner gmail account. When I post the link, it'll ask you for an email address to send you the download link (dumb, I know, but that's the way they want to do it).

EDIT: I'm just holding off on downloading the 16.2R2 version from Juniper, since both are +3GB in size.   BTW, it's weird that you get that with 17.2R1, but not 16.2R2.  I've never had that happen with any version of vMX I've tried.  I allocated the VMs 4 h/w threads and 6GB RAM between the two, though.

EDIT 2:  Could be worse.  I snagged that xrv-9K 6.2.2 image yesterday, and the .tar containing the .qcow2 file was only about 2GB, but after I extracted it, the .qcow2 image was 45GB for some bizarre reason. O_o

Edited by HermanToothrot, 24 July 2017 - 08:17 PM.


Thanked by 1 Member:
kumaragurum

#7 qianxingnan

qianxingnan

    Junior Member

  • Members
  • PipPip
  • 5 posts
  • 104 thanks

Posted 24 July 2017 - 08:50 PM

Hey Herman,

I don't how to edit in place as you did. but reply to your comments: I am running vMX 17.2 with 8192G RAM as suggested by a post. I tried running four vMXs with 6GB a couple weeks ago and some of them never had their ge-x/x/x interface names showing up.

My XRV-9K v6.1.2 was way smaller than 45GB but I have never tried v6.2.2.  If you can upload a copy I will try it out and let you know - ftp://certshare:[email protected]

Thanks,
Daniel

#8 HermanToothrot

HermanToothrot

    Member

  • Members
  • PipPip
  • 32 posts
  • 17378 thanks
  • LocationMonkey Island

Posted 24 July 2017 - 10:08 PM

You can click the Edit button on your post.  

I nuked that Xrv 9K 6.2.2 image, since it was so massive. I've still got 6.1.4, though.   I grabbed the latest from the XRv 9000 thread over in the Cisco section (I downloaded the .tar torrent).

For vMX, I just use the minimum requirements for the lab config Juniper provided.  So far, it works well, but since my current PC is based on Sandy Bridge, I can only run 2 instance in Linux (and even then, Linux starts spazzing, since it doesn't have a h/w thread for itself). When I build my new workstation, I'll have at least 16cores/32threads, and 128GB DDR4 to use, instead of the 4core/8thread and 32GB of DDR3 RAM I'm limited to.

I've never had an issue with the ge-0/0/x interfaces not showing up with any of the public trials of vMX, but you do need to wait a little while for the vFPC VM to catch up to the vCP VM.   Same is true with vQFX-10K, except it takes longer.  If you telnet into the both the vCP and vFPC VMs, you'll see the vFPC VM start listing interfaces, and then if you use "show chassis fpc pic-status" in the vCP VM, it'll only show Virtual PIC 0 online, but a quick "show interfaces ge-* terse" will list the ge-0/0/x interfaces.  I've only used 10 ge interfaces, but Junipers docs make mention of a way to increase that amount.  

Since I run the KVM versions in GNS3, I have to make 2 VMs (the last single VM vMX version was an engineering leak), and connect 1 interface from each VM to a switch (the generic built-in, OVS, CumulusVX, whatever), then connect the em1 interface of both VMs to each other, and then you can connect other topology devices to the 10 ge-0/0/x interfaces of the vFPC.   It's weird, but it's reversed with vQFX-10K.  You actually connect devices to the vRE VM, not the vPFE VM.  and we have to skip a third interface, for unknown reasons.  

Anyway, lemme know if you want the ESXi or KVM version of 16.2R2, so I can start downloading it for you. Then we can figure out how to get it to you.


EDIT:  If you're bored, or a bit curious, I wrote this guide quite a while ago, on how to use vMX and vQFX with GNS3.  I really need to update it, though.   Once I get my brain wrapped around docker, kubernetes, VMware NSX, and building tiny Qemu VMs with Alpine Linux, I'll write blog posts on those, as well as some other topics. I just don't have the time right now, to sit down an write them.

EDIT 2:  The only time I've had an issue with the ge-0/0/x interfaces not showing up with vMX was with two of the single-VM engineering leaks.  It was a simple fix to enable RIOT.  Those are missing A LOT of features, but they make some nice, lightweight L3 routers, which is the only reason I hang onto 14.1R4.8.  The rest of the time, I'll use 17.2R1.  I just wish they'd update vQFX-10K already!

Edited by HermanToothrot, 24 July 2017 - 10:19 PM.


Thanked by 1 Member:
qianxingnan

#9 qianxingnan

qianxingnan

    Junior Member

  • Members
  • PipPip
  • 5 posts
  • 104 thanks

Posted 24 July 2017 - 10:50 PM

View PostHermanToothrot, on 24 July 2017 - 10:08 PM, said:

You can click the Edit button on your post.  

I nuked that Xrv 9K 6.2.2 image, since it was so massive. I've still got 6.1.4, though.   I grabbed the latest from the XRv 9000 thread over in the Cisco section (I downloaded the .tar torrent).

For vMX, I just use the minimum requirements for the lab config Juniper provided.  So far, it works well, but since my current PC is based on Sandy Bridge, I can only run 2 instance in Linux (and even then, Linux starts spazzing, since it doesn't have a h/w thread for itself). When I build my new workstation, I'll have at least 16cores/32threads, and 128GB DDR4 to use, instead of the 4core/8thread and 32GB of DDR3 RAM I'm limited to.

I've never had an issue with the ge-0/0/x interfaces not showing up with any of the public trials of vMX, but you do need to wait a little while for the vFPC VM to catch up to the vCP VM.   Same is true with vQFX-10K, except it takes longer.  If you telnet into the both the vCP and vFPC VMs, you'll see the vFPC VM start listing interfaces, and then if you use "show chassis fpc pic-status" in the vCP VM, it'll only show Virtual PIC 0 online, but a quick "show interfaces ge-* terse" will list the ge-0/0/x interfaces.  I've only used 10 ge interfaces, but Junipers docs make mention of a way to increase that amount.  

Since I run the KVM versions in GNS3, I have to make 2 VMs (the last single VM vMX version was an engineering leak), and connect 1 interface from each VM to a switch (the generic built-in, OVS, CumulusVX, whatever), then connect the em1 interface of both VMs to each other, and then you can connect other topology devices to the 10 ge-0/0/x interfaces of the vFPC.   It's weird, but it's reversed with vQFX-10K.  You actually connect devices to the vRE VM, not the vPFE VM.  and we have to skip a third interface, for unknown reasons.  

Anyway, lemme know if you want the ESXi or KVM version of 16.2R2, so I can start downloading it for you. Then we can figure out how to get it to you.


EDIT:  If you're bored, or a bit curious, I wrote this guide quite a while ago, on how to use vMX and vQFX with GNS3.  I really need to update it, though.   Once I get my brain wrapped around docker, kubernetes, VMware NSX, and building tiny Qemu VMs with Alpine Linux, I'll write blog posts on those, as well as some other topics. I just don't have the time right now, to sit down an write them.

EDIT 2:  The only time I've had an issue with the ge-0/0/x interfaces not showing up with vMX was with two of the single-VM engineering leaks.  It was a simple fix to enable RIOT.  Those are missing A LOT of features, but they make some nice, lightweight L3 routers, which is the only reason I hang onto 14.1R4.8.  The rest of the time, I'll use 17.2R1.  I just wish they'd update vQFX-10K already!

Thank you very much for such a long reply Herman. Can you please upload the vMX KVM 16.2R2 to the FTP server I  provided in my last post? I use EVE-NG for my lab simulation and it is so easy everything can be wrapped to run in VM (50G RAM) for mixed vendor and platforms. Adding devices and bridges is just a matter of few clicks. The guide you wrote about GNS3 looks really impressive and you must have done a lot with it. I will definitely read deeper into it when I need a GNS3 box to run Juniper stuff.

I didn't know someone have uploaded a new XRV yesterday. I will download them and try them out.

Daniel

#10 HermanToothrot

HermanToothrot

    Member

  • Members
  • PipPip
  • 32 posts
  • 17378 thanks
  • LocationMonkey Island

Posted 24 July 2017 - 11:33 PM

It's uploading (I also included that free trial license, just in case you didn't have it.).  Assuming the upload doesn't fail for whatever reason, it'll likely take 3hrs to complete, due to the wildly asynch connection I get from AT&T  (25Mbit down, but 1Mbit up? WTH? I'm within spitting distance of the Uverse cab)


EDIT: BTW, if you can upload that text file to EVE, and open it with nano or something, you can use "request system add license terminal", copypasta the license into the vCP VM, and then hit Ctrl+D to apply it.  That'll give you the VMXSCALE 3 license, which enables everything.  

I'd *hate* to type that whole thing out, though.

Edited by HermanToothrot, 24 July 2017 - 11:36 PM.


#11 qianxingnan

qianxingnan

    Junior Member

  • Members
  • PipPip
  • 5 posts
  • 104 thanks

Posted 24 July 2017 - 11:56 PM

View PostHermanToothrot, on 24 July 2017 - 11:33 PM, said:

It's uploading (I also included that free trial license, just in case you didn't have it.).  Assuming the upload doesn't fail for whatever reason, it'll likely take 3hrs to complete, due to the wildly asynch connection I get from AT&T  (25Mbit down, but 1Mbit up? WTH? I'm within spitting distance of the Uverse cab)


EDIT: BTW, if you can upload that text file to EVE, and open it with nano or something, you can use "request system add license terminal", copypasta the license into the vCP VM, and then hit Ctrl+D to apply it.  That'll give you the VMXSCALE 3 license, which enables everything.  

I'd *hate* to type that whole thing out, though.

This is excellent Herman. Thank you again and really appreciate it.

Daniel

#12 HermanToothrot

HermanToothrot

    Member

  • Members
  • PipPip
  • 32 posts
  • 17378 thanks
  • LocationMonkey Island

Posted 25 July 2017 - 12:16 AM

No prob!  Happy to help when I can.  I just wish we had symmetric 1Gbps service here.. The file upload would be done already.  SecureFX is predicting it'll be done in 2hrs 30min.   I'm crossing my fingers and toes that nothing goes wrong with the upload.  :D

#13 qianxingnan

qianxingnan

    Junior Member

  • Members
  • PipPip
  • 5 posts
  • 104 thanks

Posted 25 July 2017 - 12:53 AM

No worry Herman. Upload appears to be going fine so far. It's a huge file so we need patience. Hopefully that amount transfer won't put additional cost on your bill :)

Thanks,
Daniel

#14 HermanToothrot

HermanToothrot

    Member

  • Members
  • PipPip
  • 32 posts
  • 17378 thanks
  • LocationMonkey Island

Posted 25 July 2017 - 02:23 AM

It's done.  Well, SecureFX says it's finished.  Extract the tarball, import the required images into EVE, and make sure it works. If something is screwy, then I'll try re-uploading it again tomorrow.

The md5 hash for the tarball on my end is  56a644b6f631b86b5ba081f2dcdbc4cb    (I used HashTab to calculate it, since using CertUtil in the Windows Cmd prompt is a bit annoying).




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

Organization

Community

Downloads

Test Providers

Site Info


Go to top